A multi-dimensional feature aggregation network channel estimation method based on AUV wireless communication system

Abstract

The present invention discloses a multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system, comprising the following steps: collecting channel frequency response and received signal data in the electromagnetic environment in which the AUV is located when working on the water surface to produce a data set; constructing a multi-dimensional feature aggregation network model based on a self-attention mechanism; using the data set to perform offline training on the multi-dimensional feature aggregation network; performing data preprocessing on the AUV's received signal to obtain input data; loading the offline trained network model, and inputting the input data into the network for channel estimation. The present invention adopts the above-mentioned multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system, and uses the self-attention mechanism in deep learning to build a multi-dimensional feature aggregation network model. The estimation performance is much higher than that of traditional channel estimation methods, and the network has very low spatiotemporal complexity and can be applied to AUV wireless communication systems through offline training.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communication signal processing, and in particular to a multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system. Background Art

[0002] An autonomous underwater vehicle (AUV) is an unmanned, autonomously controlled underwater vehicle. When communicating on the surface, AUVs require comprehensive predictions of the communication environment, with estimating the channel frequency response being particularly crucial. Based on the results of channel estimation, the AUV can compensate for signal fading to ensure communication reliability and adaptively adjust the transmitter's modulation and coding scheme to suit the current channel environment. Therefore, channel estimation is an essential step in AUV surface communications.

[0003] Traditional channel estimation methods primarily include pilot-based LS estimation and MMSE (Minimum Mean Square Error) estimation. LS estimation ignores the effects of noise and involves only a single division operation, making it simple and easy to implement. However, with the increasing complexity of the electromagnetic environment at sea and the increasing intensity of electromagnetic warfare in strong adversarial environments, its estimation performance falls short of the requirements for AUV surface communications. MMSE estimation is highly complex, involving two matrix inversions and requiring accurate channel statistics and noise variance. This makes MMSE estimation, despite its extremely high performance, difficult to implement in practice.

[0004] In recent years, with the development of deep learning, a growing number of researchers have utilized deep learning methods for channel estimation, achieving promising results. Existing deep learning-based channel estimation methods, such as ChannelNet and CRCENet, offer significantly higher performance than traditional methods. However, these networks are designed for channel estimation in terrestrial communications, where the communication environment is less complex and dynamic than at sea. Furthermore, these networks are computationally complex, making them difficult to adapt to AUV hardware. Summary of the Invention

[0005] The purpose of the present invention is to provide a multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system, and the time and space complexity of the network model is low. It can be applied to the AUV through offline training, achieving an estimation performance far superior to the traditional channel estimation method, thereby solving the above-mentioned technical problems existing in the prior art.

[0006] To achieve the above objectives, the present invention provides a multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system, comprising the following steps:

[0007] S1. Collect the channel frequency response and received signal data in the electromagnetic environment where the AUV is located when working on the water surface, extract the pilot signal from the received signal, and upsample the pilot signal using the interpolation method based on DPA estimation. The pilot signal is used as the input data of the training model. The channel frequency response in the collected data is used as the training label and mapped to the received signal data to form a data set.

[0008] S2. Construct a multi-dimensional feature aggregation network model based on the self-attention mechanism. The multi-dimensional feature aggregation network model FACENet based on the self-attention mechanism includes a multi-dimensional feature aggregation module, a feature processing module, and an upsampling module.

[0009] S3, in the multi-dimensional feature aggregation network model established in S2, use the data set collected in S1 for offline training to obtain a trained network model;

[0010] S4. When the AUV is actually working on the water surface, the received signal is preprocessed at the receiving end of the AUV to obtain the input data of the network model trained in S3.

[0011] S5. Input the input data obtained in S4 into the network model trained in S3 to estimate the actual channel frequency response of the AUV's environment, thereby completing channel estimation.

[0012] Preferably, in step S1, upsampling the pilot signal based on the interpolation method estimated by DPA includes the following steps:

[0013] Perform LS estimation at the pilot positions for the first OFDM symbol and the eighth OFDM symbol;

[0014] The LS estimate of the first eighth OFDM symbol is used as a virtual preamble, and the DPA estimation of the remaining OFDM symbols is performed using the virtual preamble;

[0015] Place the DPA estimation result in the corresponding position of the real channel matrix and perform linear interpolation in the OFDM symbol and subcarrier dimensions;

[0016] The interpolation result is separated into real and imaginary parts and used as input data for the model to be trained.

[0017] Preferably, in step S2, the multidimensional feature aggregation module aggregates the features of the input data in multiple dimensions, including a spatial feature aggregation block and a channel feature aggregation block; the spatial feature aggregation block extracts the spatial features of the input data from the time direction and the frequency direction; the channel feature aggregation block extracts the channel features of the input data from the channel direction.

[0018] Preferably, the spatial feature aggregation block includes a spatial self-attention block and a feed-forward network;

[0019] The spatial self-attention block consists of three parallel branches. The first branch divides the input data into 72 patches with a feature dimension of 16 along the frequency direction, and then implements a multi-head attention mechanism to extract the spatial features of the input data from the frequency dimension. The second branch divides the input data into 14 patches with a feature dimension of 16 along the time direction, and then implements a multi-head attention mechanism to extract the spatial features of the input data from the time dimension. The third branch passes the input data through a convolutional layer to extract local features. The first two branches fuse the spatial features of the frequency and time dimensions through matrix multiplication, and the result is added to the third branch to obtain the output of the spatial self-attention block.

[0020] The feedforward network uses a LayerNorm layer, two linear layers, and an activation layer to process the features extracted by the spatial self-attention block.

[0021] Preferably, the channel feature aggregation block includes a channel self-attention block and a feedforward network. The channel self-attention block consists of two parallel branches. The first branch average-pools the input data and then implements a multi-head attention mechanism to extract channel features from the channel dimension. The second branch passes the input data through a convolutional layer to extract local features. The feedforward network uses a LayerNorm layer, two linear layers, and an activation layer to process the features extracted by the two branches.

[0022] Preferably, the feature processing module further processes the extracted features, first using a convolution layer to preprocess the features extracted by the multidimensional feature aggregation module, and then using four consecutive RBs for feature processing. The output of the last RB will pass through a convolution layer to obtain the result of feature processing, and the outputs of these two convolution layers are connected through a residual structure.

[0023] Preferably, the RB is formed by cascading a convolutional layer, an activation layer whose activation function is Gelu, and a convolutional layer, and the input and output of the RB are connected by a residual structure.

[0024] Preferably, the upsampling module restores the processed features to the target size using the Pixel Shuffle upsampling method.

[0025] Preferably, in step S4, the process of data preprocessing performed by the AUV on the received signal at the receiving end is consistent with the step of up-sampling the pilot signal using the interpolation method estimated by the DPA.

[0026] Preferably, in step S5, the AUV loads the FACENet model that has been trained offline, inputs the pre-processed received data into FACENet, and estimates the actual channel frequency response of the current AUV environment through model calculation.

[0027] Therefore, the present invention adopts the above-mentioned multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system, and its technical effects are as follows:

[0028] (1) This paper uses the self-attention mechanism in deep learning to build a multi-dimensional feature aggregation network for channel estimation. The network aggregates the spatial and channel features of the input data by alternating between spatial self-attention and channel self-attention, and processes these features through a deep residual network. Considering that the channel has a strong correlation in time and frequency, spatial self-attention obtains and fuses spatial features from the directions of time and frequency respectively. This network design combines the characteristics of deep learning and OFDM signals themselves, can better extract the features of the input data, and makes the network estimation performance far superior to traditional channel estimation. Compared with other deep learning-based channel estimation schemes, it also shows better performance.

[0029] (2) The present invention uses an interpolation method based on DPA estimation to upsample the pilot signal. By taking the OFDM symbol containing the pilot as a virtual preamble, DPA estimation is performed on the remaining OFDM symbols using the virtual preamble, and linear interpolation is performed in two dimensions: OFDM symbol and subcarrier. This allows the constructed network to support flexible pilot patterns and maximize the use of pilot signals.

[0030] (3) The multi-dimensional feature aggregation network constructed by the present invention has low spatiotemporal complexity. First, the time required to complete the channel estimation of an OFDM subframe is 1.6863 ms. It can be seen that the delay caused by the calculation is allowed in the communication environment of a real AUV. In addition, the operations performed in the network model are all multiplication, division, addition and subtraction, so the implementation in hardware is not particularly difficult. Secondly, the number of model parameters of the multi-dimensional feature aggregation network is 66,746, which is significantly reduced compared to other deep learning-based channel estimation networks such as ChannlNet (the number of model parameters is 678,658) and CRCENet (the number of model parameters is 961,238). Therefore, the multi-dimensional feature aggregation network constructed by the present invention is suitable for application in AUVs.

[0031] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1Flowchart of the present invention for implementing deep learning channel estimation on AUV;

[0033] Figure 2 Schematic diagram of the pilot scheme used in the present invention;

[0034] Figure 3 A schematic diagram of the network architecture of the multi-dimensional feature aggregation network constructed by the present invention;

[0035] Figure 4 This is a schematic diagram of the network architecture of the multi-dimensional feature aggregation module in the multi-dimensional feature aggregation network;

[0036] Figure 5 Schematic diagram of the network architecture of the spatial self-attention block in the multi-dimensional feature aggregation module;

[0037] Figure 6 Schematic diagram of the network architecture of the channel self-attention block in the multi-dimensional feature aggregation module;

[0038] Figure 7 is an interpolation flow chart of the interpolation scheme based on DPA estimation;

[0039] Figure 8 It is a simulation flow chart of the channel estimation method based on the multi-dimensional feature aggregation network;

[0040] Figure 9 The following is the MSE performance simulation diagram of various channel estimation methods;

[0041] Figure 10 The BER performance simulation diagram of various channel estimation methods. DETAILED DESCRIPTION

[0042] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0043] Unless otherwise defined, technical or scientific terms used in the present invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0044] Example 1

[0045] like Figure 1 As shown, the present invention provides a multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system, comprising the following steps:

[0046] S1. Collect the channel frequency response and received signal data in the electromagnetic environment where the AUV is located when working on the water surface. Use the channel frequency response in the collected data as a training label and map it to the received signal data to form a data set.

[0047] The environment for collecting the data set is the surface of the sea area where the AUV works. Each type of sea area can be defined as a type of channel model. The data set under each type of channel model covers the received signal data and channel frequency response data under different signal-to-noise ratios (5~25dB, with 5dB as an interval). The received signal data is collected according to the AUV receiving end, and the channel frequency response data is collected according to a dedicated channel measurement device. In the collected data, the pilot signal is extracted from the received signal, and the pilot signal is upsampled using an interpolation method based on DPA estimation as the input data of the model to be trained, and the corresponding channel frequency response is used as the label. The pilot signal extracted from the received signal is distributed in a 32×2 manner in the first OFDM symbol and the eighth OFDM symbol of the 72×14 time-frequency grid, as shown in FIG. Figure 2 As shown. The pilot signal is sent along with the transmitted data by a shore-based transmitter or other AUV transmitter, and its value is 1+1j. The interpolation method based on DPA estimation is used to upsample the pilot signal, including the following steps:

[0048] S11. For the first OFDM symbol and the eighth OFDM symbol, perform LS estimation at the pilot position using formula (1):

[0049] (1);

[0050] In formula (1), is the pilot signal sent, is the received signal at the pilot, is the LS estimation result at the pilot frequency;

[0051] For example, the position of the pilot in the OFDM subframe is known to both communicating parties (shore-based and AUV, AUV and AUV) before communication. In this embodiment, the pilot is located at the 1st, 3rd, 5th, ..., 71st subcarrier positions in the 1st OFDM symbol in the OFDM subframe and the 2nd, 4th, 6th, ..., 72nd subcarrier positions in the 8th OFDM symbol. The overall pilot is distributed in a 72×14 time-frequency grid in a 32×2 grid, as shown in FIG. Figure 2 shown.

[0052] S12, the LS estimate of the first eighth OFDM symbol is used as a virtual preamble, and the DPA estimation of the remaining OFDM symbols is performed using the virtual preamble. The DPA estimation is shown in equations (2) and (3):

[0053] (2);

[0054] (3);

[0055] In formulas (2) and (3), It is received OFDM symbols, It is DPA estimation of OFDM symbols, It is equalized OFDM symbols, is the equalized OFDM symbol is mapped to the nearest constellation point.

[0056] The steps of making a virtual leader include the following:

[0057] The corresponding LS estimates are calculated using formula (1) for the 1st, 3rd, 5th, ..., 71st subcarrier positions in the 1st OFDM symbol, the channel estimation values ​​at the 2nd, 4th, 6th, ..., 72nd subcarrier positions are set to 0, and the processed 1st OFDM symbol is used as the 1st virtual preamble; similarly, the corresponding LS estimates are calculated using formula (1) for the 2nd, 4th, 6th, ..., 72nd subcarrier positions in the 8th OFDM symbol, the channel estimation values ​​at the 1st, 3rd, 5th, ..., 71st subcarrier positions are set to 0, and the processed 8th OFDM symbol is used as the 2nd virtual preamble.

[0058] S13, place the estimated result in the corresponding position of the real channel matrix, and perform linear interpolation in the OFDM symbol and subcarrier dimensions

[0059] The following steps are involved in DPA estimation at other locations:

[0060] The DPA estimation is performed on the 1st, 3rd, 5th, ..., 71st subcarrier positions on the 2nd to 14th OFDM symbols as shown in formulas (2) and (3). At this time, is the first virtual preamble produced; similarly, the DPA estimation is performed on the 2nd, 4th, 6th, ..., 72nd subcarrier positions on the 9th to 14th OFDM symbols as shown in formulas (2) and (3). At this time, This is the second virtual preamble created; for other positions involving DPA estimation, their channel estimation values ​​are all set to 0.

[0061] In step S104, linear interpolation is as shown in formula (4):

[0062] (4);

[0063] in, and is the distance interpolation point The two most recent known data points are used to perform linear interpolation in both OFDM symbol and subcarrier directions, and the completed DPA estimation results can be upsampled to the scale of a complete OFDM subframe, i.e., 72×14.

[0064] S14. Separate the real and imaginary parts of the interpolation result and use them as the third dimension of the input data. The input of the model is obtained , as the input data for the model to be trained.

[0065] S2. Construct a multi-dimensional feature aggregation network model based on self-attention mechanism suitable for AUV;

[0066] The multi-dimensional feature aggregation network model FACENet based on the self-attention mechanism includes a multi-dimensional feature aggregation module, a feature processing module and an upsampling module, such as Figure 3 The multi-dimensional feature aggregation module aggregates the features of the input data in multiple dimensions to extract the input features to the greatest extent possible; the feature processing module further processes the extracted features; and the upsampling module restores the processed features to the target size.

[0067] The multi-dimensional feature aggregation module is the core of the multi-dimensional feature aggregation network. This module aggregates the features of the input data from multiple dimensions. The module consists of a continuous spatial feature aggregation block (SFAB) and a channel feature aggregation block (CFAB), such as Figure 4 shown.

[0068] SFAB consists of a spatial self-attention block (S-SAB) and a feed-forward network (FFN), such as Figure 5As shown in Figure 2, its core lies in S-SAB. Taking into account the strong correlation between channels in time and frequency, S-SAB uses spatial self-attention to extract and fuse spatial features from both the time and frequency directions. To this end, S-SAB first leverages the concept of the Visual Transformer (VIT) to divide the input data into several patches along both the time and frequency directions, then applies self-attention to these patches. Finally, through matrix multiplication, the features extracted in both directions are fused to obtain the spatial features of the input data. Similarly, CFAB consists of a channel self-attention block (C-SAB) and a FNN, with C-SAB at its core. C-SAB average-pools the spatial features extracted by S-SAB and applies self-attention to the results along the channel direction, thereby obtaining channel-wise features. Note that C-SAB performs downsampling during the average pooling operation, allowing subsequent feature processing to be performed at a smaller scale, reducing model parameters and computational overhead. Considering that the features extracted by the self-attention mechanism are global features, we implemented a parallel convolution layer in both S-SAB and C-SAB to obtain local features. The local features obtained by convolution are added to the corresponding elements of the global features obtained by self-attention to obtain the corresponding outputs of S-SAB and C-SAB.

[0069] The spatial self-attention block consists of three parallel branches. The first branch divides the input data into 72 patches with a feature dimension of 16 along the frequency dimension and then implements a multi-head attention mechanism to extract spatial features from the frequency dimension. The second branch divides the input data into 14 patches with a feature dimension of 16 along the time dimension and then implements a multi-head attention mechanism to extract spatial features from the time dimension. The third branch passes the input data through a convolutional layer to extract local features. The first two branches fuse the spatial features from the frequency and time dimensions through matrix multiplication, and the result is added to the third branch to form the output of the spatial self-attention block. The feedforward network uses a LayerNorm layer, two linear layers, and an activation layer to process the features extracted by the spatial self-attention block.

[0070] The channel feature aggregation block consists of a channel self-attention block and a feed-forward network. Figure 6 As shown in Figure 1, the channel self-attention block consists of two parallel branches. The first branch average-pools the input data and then implements a multi-head attention mechanism to extract channel features from the channel dimension. The second branch passes the input data through a convolutional layer to extract local features. The feedforward network uses a LayerNorm layer, two linear layers, and an activation layer to process the features extracted by the two branches.

[0071] Spatial Self-Attention Block (S-SAB): Figure 5As shown, S-SAB has three branches. The first two branches are used to obtain the spatial features of the input data (from the time and frequency directions respectively), and the last branch is used to obtain the local features of the input data. , the first branch will first divide it into The feature dimension is The set of patches is The specific operation of patch division is: input data After a convolution kernel size of , the number of filters is , the convolution layer with a stride of 1, the output Refactored to The frequency self-attention layer (F-SA) will focus on Implement multi-head self-attention. First, will be mapped to a size of query( )、key( ) and value( )matrix:

[0072] (5)

[0073] in, To omit the linear mapping of the bias, the subscript Represents the frequency of its implementation of multi-head self-attention.

[0074] Next, Separate Size: The dimensions of each head are . No. Output of the head Calculated by the following formula:

[0075] (6);

[0076] Finally, the output of F-SA is obtained by connecting the outputs of all heads and performing linear mapping :

[0077] (7);

[0078] in, To omit the linear mapping of the deviation, It represents the extraction of spatial features in the frequency direction of the input data.

[0079] For input data , the second branch first divides it into The feature dimension is The set of patches is Note that the convolution kernel size of the convolution layer used in this patch division is The temporal self-attention layer (T-SA) will Implementing multi-head self-attention, the steps are exactly the same as F-SA. It is worth noting that , which is different from the F-SA The dimensions are different, the subscript Represents its implementation of temporal multi-head self-attention.

[0080] In order to implement the spatial features extracted by self-attention along the frequency Spatial features extracted with self-attention along time To perform fusion, a Reshape operation must be performed first to obtain and Then perform matrix multiplication to obtain spatial features , as shown below:

[0081] (8);

[0082] The third branch implements a convolution kernel of size , the number of filters is , a convolution layer with a step size of 1 is used to extract the local features of the input data. The result after convolution is .

[0083] Finally, through and By adding the corresponding elements, we can get the output of S-SA .

[0084] Feedforward Network (FFN): SFAB includes an LN layer, an FNN, and a residual structure in addition to S-SAB. FFN includes two linear mappings with omitted biases. And 1 activation layer GeLU. The output of S-SAB After FFN, the output of SFAB is obtained .

[0085] Channel Self-Attention Block (C-SAB): Figure 6 As shown, C-SAB has two branches. The first branch is used to obtain the channel features of the input data, and the second branch is used to obtain the local features of the input data. , the first branch first performs average pooling, and the pooling size is , the step size is The result after pooling will be reconstructed into As the input of channel self-attention (C-SA). C-SA uses formulas (5) to (7) to Processing is performed, where the resulting matrix , omitting the linear mapping of the deviation The output of C-SA The channel features of the input data are extracted. Another branch implements a convolution kernel size of , the number of filters is , the step size is The convolution layer is used to extract the local features of the input data. The result after convolution is Finally, you need to Refactored to and Add the corresponding elements to get the output of C-SAB .

[0086] Like S-SAB, C-SAB is also connected to an identical LN layer, FNN and residual structure. , the output of CFAM is .

[0087] The feature processing module further processes the extracted features. First, it uses a convolution layer to preprocess the features extracted by the multi-dimensional feature aggregation module. Then, it uses four consecutive RBs for feature processing. The output of the last RB will pass through a convolution layer to obtain the result of feature processing. The outputs of these two convolution layers are connected through a residual structure. For example, the feature processing module consists of RBs, each RB consists of two identical convolutional layers, an activation layer, and a residual structure. The convolution kernel size of the convolutional layer is , the number of filters is The activation function of the activation layer is ReLU. The residual structure adds the input and output of RB. There are two convolutional layers before and after the feature processing module, which are exactly the same as the convolutional layers in RB. For the output of module 101 After passing through the first convolution layer, it will serve as the input of the feature processing module. The output of this module passes through the second convolution layer and is added to the input through a residual structure to obtain the processed features. .

[0088] The upsampling module restores the processed features to the target size using the Pixel Shuffle upsampling method. For example, this is done using the pixel shuffle method. The upsampling module will restore the size of the real channel matrix. Before the pixel shuffle, there is a convolution kernel size of , the number of filters is The convolution layer provides sufficient depth features for pixel shuffle. After pixel shuffle, there is a convolution kernel size of , the convolution layer with 2 filters is used to make the number of output channels consistent with the number of label channels. The output of the upsampling layer is .

[0089] For the multi-dimensional feature aggregation network, we set the parameters in the network framework as follows: , , , , Finally, the total number of network training parameters is 66,746.

[0090] S3: In the multi-dimensional feature aggregation network model established in S2, offline training is performed using the dataset collected in S1 to obtain a trained network model. For example, FACENet is trained offline with hyperparameters of a learning rate of 0.001, an Adam optimizer, an L1 loss function, a batch size of 128, and a training epoch of 100 to obtain trained weights.

[0091] In S4, when the AUV is actually working on the water surface, the received signal is preprocessed at the receiving end of the AUV to obtain the input data of the network model trained in S3; the process of data preprocessing of the received signal at the receiving end of the AUV is consistent with S11 to S14.

[0092] S5. Input the input data obtained in S4 into the network model trained in S3 to estimate the actual channel frequency response of the AUV's environment, thereby completing channel estimation.

[0093] Example 2

[0094] like Figure 8 As shown, a multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system is also disclosed, including the following steps:

[0095] Step 1: Preparation of training data set

[0096] Matlab 2023a was used to simulate the maritime channel environment and generate a training dataset through simulation. Considering a SISO-OFDM system, each time slot in an OFDM subframe consists of 14 OFDM symbols, each with 72 subcarriers. Pilots are arranged in a 36×2 grid within the transmitted symbols. The system uses QPSK modulation, and the channel model is a maritime multipath channel model. The subcarrier frequency is 2.1 GHz, the carrier frequency is 15 kHz, the sampling frequency is 1.08 MHz, and the AUV speed is 50 km / h. The training dataset was generated using the aforementioned system parameters. The signal-to-noise ratio (SNR) was adjusted in 5 dB increments from 5 dB to 25 dB, with 10,000 samples generated for each SNR, for a total of 50,000 channel realizations. The training dataset has a training set to validation set ratio of 3:1.

[0097] Step 2: Offline training of the multi-dimensional feature aggregation network;

[0098] The multi-dimensional feature aggregation network FACENet was trained offline on NVIDIA GeForce RTX 3080 and Tensorflow-2.5.0 with the hyperparameters of learning rate 0.001, optimizer Adam, loss function L1 loss, batch size 128, and training cycle 100 to obtain the trained weights.

[0099] Step 3: Preparation of test data set

[0100] The system parameters used to generate the test dataset were identical to those used in the training dataset, except for the AUV's velocity. SNR was adjusted in 5dB increments from 0dB to 30dB, with 2500 samples generated for each signal-to-noise ratio, for a total of 17,500 channels. For each sample at each signal-to-noise ratio, the receiving end velocity was randomly selected from 0 to 50 km / h, meaning the Doppler frequency range for these 2500 samples was 0 to 97 Hz.

[0101] Step 4: Channel estimation through multi-dimensional feature aggregation network

[0102] Use the trained FACENet to test the test data set and get the channel estimation results. Figure 8 、 9 As shown in the figure, the minimum mean square error (MSE) and system bit error rate (BER) of channel estimation using FACENet are shown. CRCENet, Channelformer, ChannelNet and ReEsNet in the figure are other deep learning channel estimation schemes, and LMMSE, LS and DPA-interpolation are non-deep learning channel estimation schemes for comparison. Figure 9As can be seen from , FACENet has the smallest MSE under the full signal-to-noise ratio and the best estimation performance. Figure 10 As can be seen in Figure 2, FACENet has the smallest BER at all signal-to-noise ratios and the highest system reliability.

[0103] Therefore, the present invention adopts the above-mentioned multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system. The network model involved has low time and space complexity and can be applied to the AUV through offline training, achieving estimation performance far superior to traditional channel estimation methods, thereby solving the above-mentioned technical problems existing in the prior art.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-dimensional feature aggregation network channel estimation method based on AUV wireless communication system, characterized in that: The following steps are involved: S1. Collect the channel frequency response and received signal data in the electromagnetic environment where the AUV is located when working on the water surface, extract the pilot signal from the received signal, and upsample the pilot signal using the interpolation method based on DPA estimation. The pilot signal is used as the input data of the training model. The channel frequency response in the collected data is used as the training label and mapped to the received signal data to form a data set. S2. Construct a multi-dimensional feature aggregation network model based on the self-attention mechanism. The multi-dimensional feature aggregation network model FACENet based on the self-attention mechanism includes a multi-dimensional feature aggregation module, a feature processing module, and an upsampling module. The multi-dimensional feature aggregation module aggregates the features of the input data in multiple dimensions, including the spatial feature aggregation block and the channel feature aggregation block; the spatial feature aggregation block extracts the spatial features of the input data from the time direction and frequency direction; the channel feature aggregation block extracts the channel features of the input data from the channel direction; S3, in the multi-dimensional feature aggregation network model established in S2, use the data set collected in S1 for offline training to obtain a trained network model; S4. When the AUV is actually working on the water surface, the received signal is preprocessed at the receiving end of the AUV to obtain the input data of the network model trained in S3. S5. Input the input data obtained in S4 into the network model trained in S3 to estimate the actual channel frequency response of the AUV's environment, thereby completing channel estimation.

2. A multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system according to claim 1, characterized in that: In step S1, upsampling the pilot signal based on the interpolation method estimated by DPA includes the following steps: Perform LS estimation at the pilot positions for the first OFDM symbol and the eighth OFDM symbol; The LS estimate of the first eighth OFDM symbol is used as a virtual preamble, and the DPA estimation of the remaining OFDM symbols is performed using the virtual preamble; Place the DPA estimation result in the corresponding position of the real channel matrix and perform linear interpolation in the OFDM symbol and subcarrier dimensions; The interpolation result is separated into real and imaginary parts and used as input data for the model to be trained.

3. A multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system according to claim 1, characterized in that: The spatial feature aggregation block consists of a spatial self-attention block and a feed-forward network; The spatial self-attention block consists of three parallel branches. The first branch divides the input data into 72 patches with a feature dimension of 16 along the frequency direction, and then implements a multi-head attention mechanism to extract the spatial features of the input data from the frequency dimension. The second branch divides the input data into 14 patches with a feature dimension of 16 along the time direction, and then implements a multi-head attention mechanism to extract the spatial features of the input data from the time dimension; The third branch extracts local features from the input data through a convolutional layer. The first two branches fuse the spatial features of the frequency and time dimensions through matrix multiplication, and the result is added to the third branch to obtain the output of the spatial self-attention block. The feedforward network uses a LayerNorm layer, two linear layers, and an activation layer to process the features extracted by the spatial self-attention block.

4. A multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system according to claim 1, characterized in that: The channel feature aggregation block consists of a channel self-attention block and a feedforward network. The channel self-attention block consists of two parallel branches. The first branch average-pools the input data and then implements a multi-head attention mechanism to extract the channel features of the input data from the channel dimension. The second branch passes the input data through a convolutional layer to extract local features; The feedforward network uses a LayerNorm layer, two linear layers, and an activation layer to process the features extracted by the two branches.

5. The multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system according to claim 1 is characterized in that: The feature processing module further processes the extracted features. First, a convolutional layer is used to preprocess the features extracted by the multi-dimensional feature aggregation module. Then, four consecutive residual blocks RB are used for feature processing. The output of the last RB will pass through a convolutional layer to obtain the result of feature processing, and the outputs of these two convolutional layers are connected through a residual structure.

6. A multi-dimensional feature aggregation network channel estimation method based on an AUV wireless communication system according to claim 5, characterized in that: RB is composed of a convolutional layer, an activation layer with Gelu activation function, and a convolutional layer cascaded together, and the input and output of RB are connected by a residual structure.

7. The multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system according to claim 1 is characterized in that: The upsampling module restores the processed features to the target size using the Pixel Shuffle upsampling method.

8. The multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system according to claim 1 is characterized in that: In step S4, the process of data preprocessing of the received signal by the AUV at the receiving end is consistent with the step of upsampling the pilot signal using the interpolation method based on DPA estimation.

9. The multi-dimensional feature aggregation network channel estimation method based on the AUV wireless communication system according to claim 1 is characterized in that: In step S5, the AUV loads the FACENet model that has been trained offline, and inputs the pre-processed received data into FACENet. Through model calculation, the actual channel frequency response of the current AUV environment can be estimated.

Images